Add and update samples for cuda 11.0 support

2026-10-23 21:28:17 +08:00 · 2020-05-18 22:22:06 +05:30
parent 6be514679b
commit 4f6e02970b
410 changed files with 21920 additions and 3864 deletions
--- a/Samples/simpleDrvRuntime/Makefile
+++ b/Samples/simpleDrvRuntime/Makefile
@@ -72,9 +72,9 @@ endif
 # architecture
 HOST_ARCH   := $(shell uname -m)
 TARGET_ARCH ?= $(HOST_ARCH)
-ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 ppc64le armv7l))
+ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 sbsa ppc64le armv7l))
    ifneq ($(TARGET_ARCH),$(HOST_ARCH))
-        ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 ppc64le))
+        ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 sbsa ppc64le))
            TARGET_SIZE := 64
        else ifneq (,$(filter $(TARGET_ARCH),armv7l))
            TARGET_SIZE := 32
@@ -85,8 +85,17 @@ ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 ppc64le armv7l))
 else
    $(error ERROR - unsupported value $(TARGET_ARCH) for TARGET_ARCH!)
 endif
+
+# sbsa and aarch64 systems look similar. Need to differentiate them at host level for now.
+ifeq ($(HOST_ARCH),aarch64)
+    ifeq ($(CUDA_PATH)/targets/sbsa-linux,$(shell ls -1d $(CUDA_PATH)/targets/sbsa-linux))
+        HOST_ARCH := sbsa
+        TARGET_ARCH := sbsa
+    endif
+endif
+
 ifneq ($(TARGET_ARCH),$(HOST_ARCH))
-    ifeq (,$(filter $(HOST_ARCH)-$(TARGET_ARCH),aarch64-armv7l x86_64-armv7l x86_64-aarch64 x86_64-ppc64le))
+    ifeq (,$(filter $(HOST_ARCH)-$(TARGET_ARCH),aarch64-armv7l x86_64-armv7l x86_64-aarch64 x86_64-sbsa x86_64-ppc64le))
        $(error ERROR - cross compiling from $(HOST_ARCH) to $(TARGET_ARCH) is not supported!)
    endif
 endif
@@ -141,6 +150,8 @@ else ifneq ($(TARGET_ARCH),$(HOST_ARCH))
        else ifeq ($(TARGET_OS), android)
            HOST_COMPILER ?= aarch64-linux-android-clang++
        endif
+    else ifeq ($(TARGET_ARCH),sbsa)
+        HOST_COMPILER ?= aarch64-linux-gnu-g++
    else ifeq ($(TARGET_ARCH),ppc64le)
        HOST_COMPILER ?= powerpc64le-linux-gnu-g++
    endif
@@ -185,19 +196,27 @@ ifneq ($(TARGET_ARCH),$(HOST_ARCH))
                CCFLAGS += --sysroot=$(TARGET_FS)
            endif
            LDFLAGS += --sysroot=$(TARGET_FS)
-            LDFLAGS += -rpath-link=$(TARGET_FS)/lib -L $(TARGET_FS)/lib
-            LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib -L $(TARGET_FS)/usr/lib
-            LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib/aarch64-linux-gnu -L $(TARGET_FS)/usr/lib/aarch64-linux-gnu
+            LDFLAGS += -rpath-link=$(TARGET_FS)/lib -L$(TARGET_FS)/lib
+            LDFLAGS += -rpath-link=$(TARGET_FS)/lib/aarch64-linux-gnu -L$(TARGET_FS)/lib/aarch64-linux-gnu
+            LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib -L$(TARGET_FS)/usr/lib
+            LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib/aarch64-linux-gnu -L$(TARGET_FS)/usr/lib/aarch64-linux-gnu
            LDFLAGS += --unresolved-symbols=ignore-in-shared-libs
-            CCFLAGS += -isystem=$(TARGET_FS)/usr/include
-            CCFLAGS += -isystem=$(TARGET_FS)/usr/include/aarch64-linux-gnu
+            CCFLAGS += -isystem=$(TARGET_FS)/usr/include  -I$(TARGET_FS)/usr/include
+            CCFLAGS += -isystem=$(TARGET_FS)/usr/include/aarch64-linux-gnu -I$(TARGET_FS)/usr/include/aarch64-linux-gnu
+        endif
+    endif
+    ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-qnx)
+        CCFLAGS += -DWIN_INTERFACE_CUSTOM -I/usr/include/aarch64-qnx-gnu
+        LDFLAGS += -lsocket
+        LDFLAGS += -rpath=/usr/lib/aarch64-qnx-gnu -L/usr/lib/aarch64-qnx-gnu
+        ifneq ($(TARGET_FS),)
+            LDFLAGS += -rpath=$(TARGET_FS)/usr/lib -L $(TARGET_FS)/usr/lib
+            LDFLAGS += -rpath=$(TARGET_FS)/usr/libnvidia -L $(TARGET_FS)/usr/libnvidia
+        endif
+        ifdef TARGET_OVERRIDE # cuda toolkit targets override
+            NVCCFLAGS += -target-dir $(TARGET_OVERRIDE)
        endif
    endif
-endif
-
-ifeq ($(TARGET_OS),qnx)
-    CCFLAGS += -DWIN_INTERFACE_CUSTOM
-    LDFLAGS += -lsocket
 endif

 # Install directory of different arch
@@ -206,6 +225,8 @@ ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-linux)
    CUDA_INSTALL_TARGET_DIR = targets/armv7-linux-gnueabihf/
 else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-linux)
    CUDA_INSTALL_TARGET_DIR = targets/aarch64-linux/
+else ifeq ($(TARGET_ARCH)-$(TARGET_OS),sbsa-linux)
+    CUDA_INSTALL_TARGET_DIR = targets/sbsa-linux/
 else ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-android)
    CUDA_INSTALL_TARGET_DIR = targets/armv7-linux-androideabi/
 else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-android)
@@ -251,9 +272,9 @@ FATBIN_FILE := vectorAdd_kernel${TARGET_SIZE}.fatbin

 # Gencode arguments
 ifeq ($(TARGET_ARCH),$(filter $(TARGET_ARCH),armv7l aarch64))
-SMS ?= 30 35 37 50 52 60 61 70 72 75
+SMS ?= 35 37 50 52 60 61 70 72 75 80
 else
-SMS ?= 30 35 37 50 52 60 61 70 75
+SMS ?= 35 37 50 52 60 61 70 75 80
 endif

 ifeq ($(SMS),)
@@ -288,6 +309,10 @@ else
    CUDA_SEARCH_PATH ?= $(CUDA_PATH)/targets/aarch64-linux/lib/stubs
  endif

+  ifeq ($(TARGET_ARCH)-$(TARGET_OS),sbsa-linux)
+    CUDA_SEARCH_PATH ?= $(CUDA_PATH)/targets/sbsa-linux/lib/stubs
+  endif
+
  ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-android)
    CUDA_SEARCH_PATH ?= $(CUDA_PATH)/targets/armv7-linux-androideabi/lib/stubs
  endif
@@ -302,12 +327,19 @@ else

  ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-qnx)
    CUDA_SEARCH_PATH ?= $(CUDA_PATH)/targets/aarch64-qnx/lib/stubs
+    ifdef TARGET_OVERRIDE
+        CUDA_SEARCH_PATH := $(CUDA_PATH)/targets/$(TARGET_OVERRIDE)/lib/stubs
+    endif
  endif

  ifeq ($(TARGET_ARCH),ppc64le)
    CUDA_SEARCH_PATH ?= $(CUDA_PATH)/targets/ppc64le-linux/lib/stubs
  endif

+  ifeq ($(HOST_ARCH),ppc64le)
+    CUDA_SEARCH_PATH += $(CUDA_PATH)/lib64/stubs
+  endif
+
  CUDALIB ?= $(shell find -L $(CUDA_SEARCH_PATH) -maxdepth 1 -name libcuda.so 2> /dev/null)
  ifeq ("$(CUDALIB)","")
    $(info >>> WARNING - libcuda.so not found, CUDA Driver is not installed.  Please re-install the driver. <<<)
--- a/Samples/simpleDrvRuntime/README.md
+++ b/Samples/simpleDrvRuntime/README.md
@@ -10,11 +10,11 @@ CUDA Driver API, CUDA Runtime API, Vector Addition

 ## Supported SM Architectures

-[SM 3.0 ](https://developer.nvidia.com/cuda-gpus)  [SM 3.5 ](https://developer.nvidia.com/cuda-gpus)  [SM 3.7 ](https://developer.nvidia.com/cuda-gpus)  [SM 5.0 ](https://developer.nvidia.com/cuda-gpus)  [SM 5.2 ](https://developer.nvidia.com/cuda-gpus)  [SM 6.0 ](https://developer.nvidia.com/cuda-gpus)  [SM 6.1 ](https://developer.nvidia.com/cuda-gpus)  [SM 7.0 ](https://developer.nvidia.com/cuda-gpus)  [SM 7.2 ](https://developer.nvidia.com/cuda-gpus)  [SM 7.5 ](https://developer.nvidia.com/cuda-gpus)
+[SM 3.5 ](https://developer.nvidia.com/cuda-gpus)  [SM 3.7 ](https://developer.nvidia.com/cuda-gpus)  [SM 5.0 ](https://developer.nvidia.com/cuda-gpus)  [SM 5.2 ](https://developer.nvidia.com/cuda-gpus)  [SM 6.0 ](https://developer.nvidia.com/cuda-gpus)  [SM 6.1 ](https://developer.nvidia.com/cuda-gpus)  [SM 7.0 ](https://developer.nvidia.com/cuda-gpus)  [SM 7.2 ](https://developer.nvidia.com/cuda-gpus)  [SM 7.5 ](https://developer.nvidia.com/cuda-gpus)  [SM 8.0 ](https://developer.nvidia.com/cuda-gpus)

 ## Supported OSes

-Linux, Windows, MacOSX
+Linux, Windows

 ## Supported CPU Architecture

@@ -28,7 +28,7 @@ cudaMemcpy, cudaMalloc, cudaStreamCreateWithFlags

 ## Prerequisites

-Download and install the [CUDA Toolkit 10.2](https://developer.nvidia.com/cuda-downloads) for your corresponding platform.
+Download and install the [CUDA Toolkit 11.0](https://developer.nvidia.com/cuda-downloads) for your corresponding platform.

 ## Build and Run

@@ -67,29 +67,5 @@ The samples makefiles can take advantage of certain options:
    $ make HOST_COMPILER=g++
 ```

-### Mac
-The Mac samples are built using makefiles. To use the makefiles, change directory into the sample directory you wish to build, and run make:
-```
-$ cd <sample_dir>
-$ make
-```
-
-The samples makefiles can take advantage of certain options:
-
-*  **dbg=1** - build with debug symbols
-    ```
-    $ make dbg=1
-    ```
-
-*  **SMS="A B ..."** - override the SM architectures for which the sample will be built, where "A B ..." is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use SMS="50 60".
-    ```
-    $ make SMS="A B ..."
-    ```
-
-*  **HOST_COMPILER=<host_compiler>** - override the default clang host compiler. See the [Mac Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-mac-os-x/index.html#system-requirements) for a list of supported host compilers.
-    ```
-    $ make HOST_COMPILER=clang
-    ```
-
 ## References (for more details)

--- a/Samples/simpleDrvRuntime/simpleDrvRuntime_vs2012.vcxproj
+++ b/Samples/simpleDrvRuntime/simpleDrvRuntime_vs2012.vcxproj
@@ -62,7 +62,7 @@
      <OutputFile>$(OutDir)/simpleDrvRuntime.exe</OutputFile>
    </Link>
    <CudaCompile>
-      <CodeGeneration>compute_30,sm_30;compute_35,sm_35;compute_37,sm_37;compute_50,sm_50;compute_52,sm_52;compute_60,sm_60;compute_61,sm_61;compute_70,sm_70;compute_75,sm_75;</CodeGeneration>
+      <CodeGeneration>compute_35,sm_35;compute_37,sm_37;compute_50,sm_50;compute_52,sm_52;compute_60,sm_60;compute_61,sm_61;compute_70,sm_70;compute_75,sm_75;compute_80,sm_80;</CodeGeneration>
      <AdditionalOptions>-Xcompiler "/wd 4819" %(AdditionalOptions)</AdditionalOptions>
      <Include>./;../../Common</Include>
      <Defines>WIN32</Defines>
--- a/Samples/simpleDrvRuntime/simpleDrvRuntime_vs2013.vcxproj
+++ b/Samples/simpleDrvRuntime/simpleDrvRuntime_vs2013.vcxproj
@@ -62,7 +62,7 @@
      <OutputFile>$(OutDir)/simpleDrvRuntime.exe</OutputFile>
    </Link>
    <CudaCompile>
-      <CodeGeneration>compute_30,sm_30;compute_35,sm_35;compute_37,sm_37;compute_50,sm_50;compute_52,sm_52;compute_60,sm_60;compute_61,sm_61;compute_70,sm_70;compute_75,sm_75;</CodeGeneration>
+      <CodeGeneration>compute_35,sm_35;compute_37,sm_37;compute_50,sm_50;compute_52,sm_52;compute_60,sm_60;compute_61,sm_61;compute_70,sm_70;compute_75,sm_75;compute_80,sm_80;</CodeGeneration>
      <AdditionalOptions>-Xcompiler "/wd 4819" %(AdditionalOptions)</AdditionalOptions>
      <Include>./;../../Common</Include>
      <Defines>WIN32</Defines>
--- a/Samples/simpleDrvRuntime/simpleDrvRuntime_vs2015.vcxproj
+++ b/Samples/simpleDrvRuntime/simpleDrvRuntime_vs2015.vcxproj
@@ -62,7 +62,7 @@
      <OutputFile>$(OutDir)/simpleDrvRuntime.exe</OutputFile>
    </Link>
    <CudaCompile>
-      <CodeGeneration>compute_30,sm_30;compute_35,sm_35;compute_37,sm_37;compute_50,sm_50;compute_52,sm_52;compute_60,sm_60;compute_61,sm_61;compute_70,sm_70;compute_75,sm_75;</CodeGeneration>
+      <CodeGeneration>compute_35,sm_35;compute_37,sm_37;compute_50,sm_50;compute_52,sm_52;compute_60,sm_60;compute_61,sm_61;compute_70,sm_70;compute_75,sm_75;compute_80,sm_80;</CodeGeneration>
      <AdditionalOptions>-Xcompiler "/wd 4819" %(AdditionalOptions)</AdditionalOptions>
      <Include>./;../../Common</Include>
      <Defines>WIN32</Defines>
--- a/Samples/simpleDrvRuntime/simpleDrvRuntime_vs2017.vcxproj
+++ b/Samples/simpleDrvRuntime/simpleDrvRuntime_vs2017.vcxproj
@@ -67,7 +67,7 @@
      <OutputFile>$(OutDir)/simpleDrvRuntime.exe</OutputFile>
    </Link>
    <CudaCompile>
-      <CodeGeneration>compute_30,sm_30;compute_35,sm_35;compute_37,sm_37;compute_50,sm_50;compute_52,sm_52;compute_60,sm_60;compute_61,sm_61;compute_70,sm_70;compute_75,sm_75;</CodeGeneration>
+      <CodeGeneration>compute_35,sm_35;compute_37,sm_37;compute_50,sm_50;compute_52,sm_52;compute_60,sm_60;compute_61,sm_61;compute_70,sm_70;compute_75,sm_75;compute_80,sm_80;</CodeGeneration>
      <AdditionalOptions>-Xcompiler "/wd 4819" %(AdditionalOptions)</AdditionalOptions>
      <Include>./;../../Common</Include>
      <Defines>WIN32</Defines>
--- a/Samples/simpleDrvRuntime/simpleDrvRuntime_vs2019.vcxproj
+++ b/Samples/simpleDrvRuntime/simpleDrvRuntime_vs2019.vcxproj
@@ -63,7 +63,7 @@
      <OutputFile>$(OutDir)/simpleDrvRuntime.exe</OutputFile>
    </Link>
    <CudaCompile>
-      <CodeGeneration>compute_30,sm_30;compute_35,sm_35;compute_37,sm_37;compute_50,sm_50;compute_52,sm_52;compute_60,sm_60;compute_61,sm_61;compute_70,sm_70;compute_75,sm_75;</CodeGeneration>
+      <CodeGeneration>compute_35,sm_35;compute_37,sm_37;compute_50,sm_50;compute_52,sm_52;compute_60,sm_60;compute_61,sm_61;compute_70,sm_70;compute_75,sm_75;compute_80,sm_80;</CodeGeneration>
      <AdditionalOptions>-Xcompiler "/wd 4819" %(AdditionalOptions)</AdditionalOptions>
      <Include>./;../../Common</Include>
      <Defines>WIN32</Defines>